|
|
Received: 27 May 2016 Revised: 11 October 2016 Accepted: 19 October 2016
DOI: 10.1002/ece3.2606
ORIGINAL RESEARCH
Co-­occurrence patterns in a diverse arboreal ant community
are explained more by competition than habitat requirements
Flávio Camarota1,2 | Scott Powell2 | Adriano S. Melo3 | Galen Priest4 | Robert J. Marquis4 | Heraldo L. Vasconcelos1
1
Instituto de Biologia, Universidade Federal de
Uberlândia, Uberlândia, Brazil
Abstract
2
A major goal of community ecology is to identify the patterns of species associations
Department of Biological Sciences, The
George Washington University, Washington,
DC, USA
3
Departamento de Ecologia, Universidade
Federal de Goiás, Goiânia, Brazil
4
Department of Biology and the Whitney
R. Harris World Ecology Center, University of
Missouri – St. Louis, St. Louis, MO, USA
Correspondence
Flávio Camarota, Department of Biological
Sciences, The George Washington University,
Washington, DC, USA.
E-mail: [email protected]
Funding information
National Science Foundation, Grant/Award
Number: DEB 0842144; Brazilian Council
for Research and Scientific Development,
Grant/Award Number: 478107/2012-9
and 457407/2012-3; National Science
Foundation, Grant/Award Number: IOS
0841756; George Washington University
and the processes that shape them. Arboreal ants are extremely diverse and abundant,
making them an interesting and valuable group for tackling this issue. Numerous studies have used observational data of species co-occurrence patterns to infer underlying
assembly processes, but the complexity of these communities has resulted in few solid
conclusions. This study takes advantage of an observational dataset that is unusually
well-structured with respect to habitat attributes (tree species, tree sizes, and vegetation structure), to disentangle different factors influencing community organization. In
particular, this study assesses the potential role of interspecific competition and habitat selection on the distribution patterns of an arboreal ant community by incorporating habitat attributes into the co-occurrence analyses. These findings are then
contrasted against species traits, to explore functional explanations for the identified
community patterns. We ran a suite of null models, first accounting only for the species incidence in the community and later incorporating habitat attributes in the null
models. We performed analyses with all the species in the community and then with
only the most common species using both a matrix-level approach and a pairwise-level
approach. The co-occurrence patterns did not differ from randomness in the matrixlevel approach accounting for all ant species in the community. However, a segregated
pattern was detected for the most common ant species. Moreover, with the pairwise
approach, we found a significant number of negative and positive pairs of species associations. Most of the segregated associations appear to be explained by competitive
interactions between species, not habitat affiliations. This was supported by comparisons of species traits for significantly associated pairs. These results suggest that competition is the most important influence on the distribution patterns of arboreal ants
within the focal community. Habitat attributes, in contrast, showed no significant influence on the matrix-wide results and affected only a few associations. In addition,
the segregated pairs shared more biological characteristic in common than the aggregated and random ones.
KEYWORDS
assembly rules, Brazil, canopy, Cerrado, community assembly, niche, species coexistence
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium,
provided the original work is properly cited.
© 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd.
Ecology and Evolution 2016; 1–12
www.ecolevol.org | 1
|
CAMAROTA et al.
2 1 | INTRODUCTION
whole-­community level, which ignores potentially important interspecific interactions (Diamond & Gilpin, 1982) and weakens inferences
A fundamental goal of community ecology is to understand the mech-
about the factors that are responsible for the observed patterns
anisms allowing multispecies coexistence in biological communities
(Sanderson, 2004; Veech, 2014). This is because we need to know
(Agrawal et al., 2007; Sutherland et al., 2013). To achieve this goal, it is
the negatively or positively associated species pairs to compare the
important to assess how species are distributed within communities,
biological traits and habitat affinities that may explain the observed
and what factors and processes generate these patterns (Chesson,
association patterns (Veech, 2014). To learn more from species co-­
2000). Niche-­based processes are often offered as an explanation
occurrence data, it is therefore necessary to use null model analyses
for species distributions within communities, with the general ex-
that incorporate potentially important environmental influences, and
pectation that species must differ in resource use to coexist (Chase
contrast whole-­community versus pairwise patterns (Gotelli, 2001;
& Leibold, 2003; Levine & HilleRisLambers, 2009). Therefore, commu-
Ulrich & Gotelli, 2013).
nity assembly may be shaped strongly by biotic interactions among
Ants are abundant, highly diverse, and ecologically important
species, and particularly by interspecific competition (Chase & Leibold,
in tropical systems. Moreover, ants engage in obvious and often in-
2003; Hutchinson, 1959; MacArthur & Levins, 1967; Schoener, 1974).
tense competitive interactions (Hölldobler & Wilson, 1990; Lach,
Nevertheless, the role of interspecific competition in shaping commu-
Parr, & Abbot, 2010). These attributes, combined with the relative
nities is a complex issue that remains under active investigation and
ease with which communities can be sampled, make them especially
debate.
well suited for use in co-­occurrence analyses. Not surprisingly, many
In addressing the role of competition in community assembly, much
studies have evaluated co-­occurrence patterns in ant communities
work has focused on detecting the signature of competition in species
(Gotelli & McCabe, 2002), particularly in tropical arboreal ant commu-
co-­occurrence patterns within communities (Gotelli & McCabe, 2002).
nities (Dejean et al., 2010; Pfeiffer, Cheng Tuck, & Chong Lay, 2008;
The central assumption of these analyses is that if competitive interac-
Sanders, Crutsinger, Dunn, Majer, & Delabie, 2007). Indeed, arboreal
tions scale up to shape community-­level species distribution patterns,
ant communities have been seen as a classic example of a nonran-
nonrandom patterns of localized species co-­occurrence should be ob-
dom pattern of species co-­occurrence (Blüthgen & Stork, 2007). More
servable within a community (Connor & Simberloff, 1979; Diamond,
specifically, it has been hypothesized that these communities have a
1975). These analyses are based on statistical null models, which un-
“mosaic distribution,” where two or more dominant ant species main-
like traditional statistical tests, explicitly assess the importance of sto-
tain mutually exclusive territories, and each coexists with a specific set
chastic processes on community organization (Gotelli & Graves, 1996;
of subordinate and subdominant ant species (Jackson, 1984; Leston,
Gotelli & Ulrich, 2012). Moreover, a null model approach allows the in-
1970). This mosaic distribution pattern is proposed as a repeatable,
clusion of constraints to preserve features of the empirical data, which
specific outcome of intense competition between dominant species
cannot be done easily in parametrized statistical tests (Gotelli & Ulrich,
over foraging territories (Blüthgen & Feldhaar, 2010; Blüthgen & Stork,
2012). These kind of analyses are particularly valuable because they
2007). Nevertheless, the use of co-­occurrence analyses to address
facilitate the use of large biodiversity-­survey datasets to address the
the role of competition in structuring arboreal ant communities, and
underlying processes of community assembly (Gotelli, 2000; Gotelli &
especially the specific case of the “ant mosaic hypothesis,” has been
Graves, 1996). However, it is important to note that nonrandom pat-
criticized heavily. This criticism was based on the shortcomings of an-
terns of species co-­occurrence are not necessarily the result of com-
alytical methodologies for considering other influences on nonrandom
petitive interactions (Connor & Simberloff, 1979, 1984; Peres-­Neto,
co-­occurrence patterns (Gotelli, 2001; Ribas & Schoereder, 2002).
Olden, & Jackson, 2001). Other factors not directly related to compe-
To avoid past issues with co-­occurrence analyses of arboreal ant
tition, such as dispersal ability and habitat requirements, can produce
communities, it is necessary to analyze a suite of factors that may
similar patterns (Azeria, Ibarzabal, & Hébert, 2012; Sanderson, 2004;
produce nonrandom co-­occurrence patterns, but that are not directly
Sfenthourakis, Tzanatos, & Giokas, 2006). It is also possible that in-
related to competition. Environmental factors are a particularly im-
terspecific niche similarities can be more important in shaping com-
portant focus in this regard, because they are often reflected in the
munity assembly than niche differences, a process known as “niche
habitat use of different species. For arboreal ants, a single tree is a
filtering” (Fowler, Lessard, & Sanders, 2014).
distinct habitat patch, and tree size, tree species, and the surround-
Given the various potential influences on community assembly,
ing vegetation can be considered key habitat attributes (Pacheco &
co-­occurrence analyses become more valuable if they can differenti-
Vasconcelos, 2012; Ribas, Schoereder, Pic, & Soares, 2003). In fact,
ate between the signature of competition and other factors (Connor
trees of different sizes typically host different abundance and rich-
& Simberloff, 1979; Sfenthourakis et al., 2006). Once again, a null
ness of arboreal ant species (Campos, Vasconcelos, Ribeiro, Neves, &
model approach is ideal to tackle this issue, as its flexibility allows
Soares, 2006; Koch, Camarota, & Vasconcelos, 2016; Powell, Costa,
the explicit incorporation of the variables of interest in the analysis
Lopes, & Vasconcelos, 2011). Similarly, some ant species may nest
(Gotelli & Graves, 1996). Nevertheless, co-­occurrence analyses are
only on specific tree species (Klimes et al., 2012; Sanders, Gotelli
typically applied to large datasets that lack the sampling structure to
et al., 2007), and the structure of the surrounding vegetation can also
address other influences on co-­occurrence patterns (Gotelli, 2001).
have a significant impact of species richness (Powell et al., 2011). With
Additionally, co-­occurrence patterns are usually assessed only at the
all these aspects considered, it is possible that a nonrandom pattern
|
3
CAMAROTA et al.
of species associations may result from distinct habitat use and not
and Kielmeyeria coriacea Mart. & Zucc. (Clusiaceae). These species are
interspecific competition, as it is often claimed. Nevertheless, most
common in the study area, and in the Cerrado as a whole (Moreno,
arboreal ant datasets are not structured in a way that allows for co-­
2005). Half of the trees were sampled in areas of cerrado ralo (open
occurrence analyses to systematically address the potential influence
savannas hereafter) and the other half in areas of cerrado sensu stricto
of environmental factors on community assembly.
(closed savannas hereafter). Each tree species had the same number
In this study, we assess the extent to which species coexistence
of individuals sampled in each vegetation type. Trees from each spe-
patterns in arboreal ants can be explained by interspecific competition
cies were classified as “small,” “medium,” or “large” based on trunk
or by habitat associations. We further address whether the biological
diameter 10 cm above the soil surface. Small trees had a diameter up
characteristics of the most common ant species, such as nesting pref-
to 11.9 cm (60 trees), medium trees were between 12 and 21.9 cm
erences and recruitment strategies, can explain the patterns we un-
in diameter (121 trees), and large trees had a diameter of 22–36 cm
cover. We achieved this with data from a Neotropical savanna, where
(59 trees).
the natural habitat allowed for an extensive biodiversity survey struc-
We used baited arboreal pitfall traps for the ant sampling, follow-
tured by tree size, tree species, and vegetation structure, and where
ing the procedure of Powell et al. (2011). Briefly, the number of traps
the biological characteristics of ant species were quantifiable. To ana-
per tree ranged from 4 to 10, according to the size of the tree. On
lyze these data, we used a suite of null models that first account only
each tree, half of the traps contained honey diluted in water (1:7) and
for the species incidence in the community, and then explicitly incor-
the other half contained human urine (diluted 1:2 in water). A small
porate habitat attributes. We further performed all analyses first with
quantity of detergent was included in the bait liquids to break the sur-
all species in the community, and then with only the most common
face tension and thus improve the killing efficiency. Each pitfall trap
species. Moreover, as the community-­wide analyses can hide relevant
consisted of approximately 20 ml of bait liquid in an 80-­ml plastic cup
species interactions (Veech, 2014), we also used pairwise analysis to
(6 cm height and 5 cm diameter). The traps were distributed through-
search for co-­occurrence patterns of different species pairs. Following
out the crown of each focal tree, using wire to hold the rim horizontal
this, we assessed similarities in key biological properties of the most
and touching a branch. Traps were left on the trees for 48 h for the
common ant species in order to interpret the observed pairwise pat-
collection of both diurnal and nocturnal ant species. All sampling was
terns. We used these analyses to address three main questions: (1) do
completed over a period of 10 days, using a stratified sampling proce-
co-­occurrence patterns in our focal arboreal ant community deviate
dure of four trees of different sizes per species on each day, for a total
significantly from that expected under a process of random community
of 24 trees sampled each day. The contents of all traps were sorted
assembly? (2) to what extent do these co-­occurrence patterns change
and identified to species and morphospecies in the laboratory. All ma-
when we account for potentially important habitat attributes? (3) to
terial was compared against the existing ant collection at the Federal
what extent can the observed co-­occurrence patterns be explained by
University of Uberlândia, which is built from extensive collections at
differences or similarities in key traits of the species involved?
many Cerrado sites. Voucher specimens of all species/morphospecies are held at the Zoological Collection of the Federal University of
2 | MATERIALS AND METHODS
2.1 | Study area
This study was conducted at the Reserva Ecológica do Panga, a 404-­
ha reserve located 35 km south of Uberlândia, Minas Gerais, Brazil
Uberlândia in Brazil.
2.3 | Null model analyses
2.3.1 | Co-­occurrence metrics
(19°10′ S, 48°23′ W). The reserve is located in the Neotropical sa-
To assess species co-­occurrence pattern, we used a standardized ver-
vannas of central Brazil (Cerrado hereafter). Cerrado is characterized
sion of the C-­score of Stone and Robert (1990) (St. C-­score), which
by a vegetation mosaic dominated by savannas of variable structure
corrects for the differences in species incidence within the commu-
(Oliveira-­Filho & Ratter, 2002). This study was conducted in two kinds
nity (Azeria et al., 2009). The C-­score measures the mean number
of savannas with marked differences in their vegetation structure: the
of checkerboard units between all possible pairs of species in a data
cerrado ralo, with scattered distributed trees and higher grass cover,
matrix (Stone & Robert, 1990). It can, however, also be used for analy-
and the cerrado sensu stricto, with a higher tree density and lower
sis between a single pair of species. When compared to other indices
grass cover.
used to assess co-­occurrence, the C-­score has a smaller probability of
type I and II errors (Gotelli, 2000). The number of checkerboard units
2.2 | Ant sampling
We collected ants on a total of 240 trees, with even sampling of 40
(CU) for each species pair in the standardized version of the index
is: SCU = (ri − A)(rj − A)/(ri*rj), where A is the number of shared sites
(trees) and ri and rj are the number of trees on which each species i
trees for each of the following six tree species: Caryocar brasiliense
and j, respectively, are found (Stone & Robert, 1990). We also used
Cambess. (Caryocaraceae), Qualea grandiflora Mart. (Vochysiaceae),
the Sørensen index (SOR) (Dice, 1945) to measure the mean number
Stryphnodendron polyphyllum Mart. (Fabaceae), Sclerolobium aureum
of shared sites (i.e., trees in our study) between the different pairs
(Tul.) Benth. (Caesalpiniaceae), Machaerium opacum Vogel (Fabaceae),
of species: The index is calculated as SOR = 2A/(2A + B + C) where A
|
CAMAROTA et al.
4 represents the number of shared sites between species r and j, while B
In the case where one of the two indexes (e.g., St. C-­score) is sig-
represents the number of sites where species r is present and species
nificantly different from the null distribution under both the uncon-
j is absent, and C is the converse of B.
strained and habitat-­constrained models, this would suggest that
The magnitude of the standardized C-­score and the Sørensen
some interspecific interaction is indeed helping to explain the ob-
index depend on the frequency of occurrence of the species, preclud-
served pattern. On the other hand, if the index is significant under the
ing a direct interpretation of the observed values. Accordingly, ob-
unconstrained model but not under the habitat-­constrained model, it
served values of these statistics can be assessed by comparing them
suggests that the habitat selection is more important for explaining
to a distribution generated by a null model. This is done by calculating
the observed patterns than competition (Azeria et al., 2012; Peres-­
a standardized effect size (SES), so that results are comparable across
Neto et al., 2001). Another possible outcome is that species pairs that
communities, pairs of species, and even with other studies. The SES
are not significant under unconstrained analyses can be significant
is calculated as follows: SES = (observed index value − mean of simu-
under habitat-­constrained null models. This may indicate that there is
lated index values)/standard deviation of simulated index values. The
a negative association within a shared habitat (for segregated pairs) or
significance of the observed SES values was computed as the propor-
that the differences in species affinity for a given habitat could have
tion of simulated values equal to or more extreme than the observed
masked an otherwise positive interaction (for aggregated pairs) (Azeria
value (Gotelli, 2000). SES values greater than 1.98 (p < .05) indicate a
et al., 2012).
segregated distribution under the St. C-­Score and aggregated distribution under the SOR and vice-­versa when SES values are <−1.98.
We assessed species co-­occurrences using the fixed–fixed (FF)
2.3.3 | Matrix-­level versus pairwise-­level approaches
null model, where the frequencies of occurrence (fixed columns)
Most of the methods used for co-­occurrence analyses can be classified
and the original species richness on each tree are maintained (fixed
as either “matrix” or “pairwise” approaches. The matrix approach calcu-
rows). Thus, differences among sites are maintained, but the species
lates the co-­occurrence metrics as a property of the whole presence/
occurrences are random with respect to one another, which makes
absence species matrix (Gotelli, 2000; Pitta, Giokas, & Sfenthourakis,
it appropriate for detecting patterns of species interactions (Gotelli,
2012). We first calculated null models for whole matrices and then
2000). Furthermore, the FF null model is not vulnerable to type I error,
performed pairwise analyses to assess the co-­occurrence for each spe-
has power in the face of noisy data, and measures a pattern of co-­
cies pair separately. This was done to determine whether each possible
occurrence consistent with competitive exclusion (Gotelli, 2000).
pair in the community has an aggregated, segregated, or random co-­
occurrence pattern (Gotelli & Ulrich, 2012; Pitta et al., 2012; Veech,
2014). We present data for the two indices considered (St. C-­score
2.3.2 | Unconstrained versus habitat-­constrained
null models
and Sorensen) on the matrix-­level analyses. However, for the pairwise
We further implemented two different classes of the FF null model: one
were more numerous and qualitatively identical across both indices.
analyses we present results for only the St. C-­score, as the outcomes
that takes into account only aspects of the species occurrence and site
For the whole matrix approach, we conducted analyses first includ-
species richness of the original matrix (“unconstrained null models”), and
ing all sampled ant species (75 spp.) and then only the 14 most com-
one that incorporates the species specificity for a given habitat on the
mon ant species. For the pairwise approach, we conducted analyses
species null distribution (“habitat-­constrained null models”) (Azeria et al.,
only with the 14 most common ant species. Restricting the pairwise
2012; Sfenthourakis et al., 2006). The unconstrained model was con-
analyses to the common species avoids the loss of statistical power
trasted against three different habitat-­constrained models, each using
associated with including species that were rare. These 14 ant species
a different environmental variable. The environmental variables used in
were chosen because they were found in at least 20 of the sampled
the habitat-­constrained models were as follows: tree species (Caryocar
trees. These 14 ant species also included the most abundant species
brasiliense, Qualea grandiflora, Stryphnodendron polyphyllum, Sclerolobium
in the community, representing 93% of the total number of species
aureum, Machaerium opacum, Kielmeyeria coriacea), tree size (small, me-
incidences across all sampled trees.
dium, large), and vegetation structure (open or closed). These three
different environmental variables are related to the host tree characteristics and were chosen for their potential impact on ant species richness
2.3.4 | Matrix randomizations
and composition (Djiéto-­Lordon, Dejean, Gibernau, Hossaert-­McKey, &
We ran 5,000 randomizations for each null model. Randomizations
McKey, 2004; Klimes et al., 2012; Powell et al., 2011).
were done using the function “commsimulator” in R environment ver-
The constrained null models differ from their unconstrained coun-
sion 3.3.3 (R Development Core Team 2014), available in the “vegan”
terparts by restricting the randomization of occurrence values to the
package (Oksanen, Kindt, Legendre, O’Hara, & Stevens, 2013). The
same level of the constraining factor. For instance, an ant species re-
simulated communities were obtained using the “quasiswap” algo-
cord found on the level “Caryocar brasiliense” of the factor tree species
rithm, where each simulation uses the original matrix and not the
will be randomized among individuals of “Caryocar brasiliense.” It will
previous randomized matrices (Gotelli & Entsminger, 2001, 2003).
not be assigned, during randomizations to generate simulated commu-
We wrote R routines for the pairwise analyses, and the habitat-­
nities, to a different host plant species.
constrained analyses.
|
5
CAMAROTA et al.
2.4 | Species–habitat associations
We assessed whether ant species were associated with a particu-
“small,” between one and five workers, “medium,” from 6 to 10 workers, “large,” from 11 to 25 workers, and “very large,” with more than
25 workers.
lar tree species, tree size, or trees in the open or closed areas using
Indicator Species Analysis (Dufrene & Legendre, 1997). This analysis
estimates the strength of the association of different species with distinct groups of sites (Cáceres & Legendre, 2009). We used the IndVal
2.5.3 | Differences between segregated and
aggregated pairs
program of Dufrene and Legendre (1997) and assessed the signifi-
To assess differences in dissimilarity in traits between species that
cance of the indicator values for each species using Monte Carlo′s
formed segregated and aggregated pairs, we performed a logistic re-
permutation tests with 5,000 randomizations.
gression, with the dissimilarity (Sørensen) between the species pairs
as the predictor variable and the kind of association (1 = segregated,
2.5 | Nesting and foraging ecology of the most
common ant species
2.5.1 | Ant nesting ecology
To assess the nesting ecology of the 14 most common species, we
opened and measured stems of different sizes belonging to 20 trees
of each of the six tree species studied. In each tree, we removed six
branches of a given base diameter size and then dissected each base
stem and all subsequent stems up to the terminal tips. We set a base
0 = aggregated) as the response variable. For this analysis, we excluded those species pairs that were determined by habitat variables
(one segregated and two aggregated pairs).
3 | RESULTS
3.1 | Description of the community and ecology of
the most common species
Overall, we collected 75 ant species from 17 genera (Table S1). Most
diameter of approximately 2 cm for eight trees and a base diameter of
ant species were very low in frequency in the community, with only
5 cm for 12 of the sampled trees. For each ant nest found, we meas-
14 species occurring on at least 20 of the 240 sampled trees. The
ured the mean length of the nesting cavity. Nests were further clas-
most frequent species included five species of Camponotus, four spe-
sified as being in dead or live stems, with cavities that were small,
cies of Pseudomyrmex, Azteca sp. 1, Cephalotes pusillus, Crematogaster
medium, large, or very large in length, or simply under the tree bark.
ampla, Solenopsis sp. 1, and Tapinoma sp. 1. The nesting and forag-
Small cavities were up to 5 cm of length, medium cavities from 5.1 to
ing ecology of these species are presented in Table 1. Most species
25 cm, large cavities those from 25.1 to 50 cm, and very large cavities
of Pseudomyrmex were strictly diurnal, recruited very few workers to
those with more than 51 cm of length. As the percentage of the tree
baits, and nested exclusively in small branches. Among the Camponotus
occupied by an ant species can be important to define its dominance
species, Ca. melanoticus was only found nesting in cavities located in
status, we also calculated the occupation rates of the available cavi-
live branches of medium size, Ca. sericeiventris was only in cavities lo-
ties by the different ant species. For this, we summed the total length
cated in large and very large sized live branches, whereas the remain-
of used cavities by each species and divided by the total length of
ing three species nested in branches of small, medium, and large size
available cavities on the trees where each species was found. Those
(Table 1). Three of the Camponotus species were strictly nocturnal,
species that occupied more than 25% of the available cavities were
whereas the remaining two forage during the day and the night. Five
considered as having “extensive cavity use.”
of the 14 species had extensive cavity use, but only three of those
(Azteca sp. 1, Crematogaster ampla, and Cephalotes pusillus) had large
2.5.2 | Ant activity schedule and recruitment to baits
colonies whose nests are located in live and dead branches of any size.
With the exception of Azteca sp. 1, which was found mostly on
We observed ant activity at baits on 175 trees during the day (be-
S. aureum trees (Table 2), none of the remaining species had a signifi-
tween 08:00 and 12:00) and a subset of 44 of these trees during the
cant association with a given tree species (Indicator Species Analysis,
night (between 19:00 and 23:00). These were different trees from
Table 2). Similarly, with the exception of Ca. sericeiventris, which was
those used in the ant survey (above), and the baits were solid pieces
mostly found on trees located in the closed savanna habitats, none
of sardine. Observations were made on 8–10 trees each day until ob-
of the remaining ant species showed a significant association with a
servations had been made on all trees. For each tree, activity at the
given vegetation structure. However, four of the 14 species showed
baits was observed within ten minutes of the baits being placed on the
significant associations with trees of a given size (Table 2).
tree, and then again after one, two, and three hours. Ant species were
classified as “diurnal” if they were only or mostly seen foraging during
the day, “nocturnal,” if they were seen only or mostly during the night
and “both” if they were seen foraging both day and night. We also
recorded the maximum number of ants of each species present on a
3.2 | Null model analyses
3.2.1 | Matrix-­level co-­occurrence analyses
given bait across the four observations periods. Recruitment strength
The analyses with all 75 ant species in the community showed random
was then summarized into four different categories for the analyses:
patterns under the unconstrained models for the two indexes used
x
x
x
x
x
x
x
x
x
x
x
x
x
Cr. ampla
P. curacaensis
P. elongatus
P. gracilis
P. urbanus
Solenopsis sp. 1
Tapinoma sp. 1
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
On the Recruitment size, S = small, M = medium, L = large, and XL = very large recruitment rates.
x
x
x
x
Ce. pusillus
x
x
x
x
x
x
Ca. bonariensis
Ca. senex
x
x
Ca. atriceps
x
x
x
x
x
x
x
x
x
x
x
x
Live very
large
branches
x
x
x
x
x
Dead very
large
branches
x
Under
bark
Yes
Extensive
cavity use
x
No
Yes
No
No
No
No
Yes
Yes
No
No
No
Yes
x
Dead large
branches
x
x
Live large
branches
Ca. sericeiventris
x
Dead
medium
branches
No
Azteca sp. 1
Live
medium
branches
x
x
Ant species
Dead small
branches
Ca. melanoticus
Live small
branches
Nesting ecology
Species ecological characteristics
T A B L E 1 Ecological characteristics of the 14 most common ant species in the study area
Both
Both
Diurnal
Diurnal
Both
Diurnal
Both
Both
Diurnal
Nocturnal
Nocturnal
Both
Nocturnal
Both
Activity
period
Other
XL
XL
S
S
S
S
XL
L
M
L
L
M
M
XL
Recruit.
6 |
CAMAROTA et al.
|
7
CAMAROTA et al.
T A B L E 2 Species–habitat association,
with the IndVal results of the 14 most
common arboreal ants on our study area
Ant species
Tree species
Vegetation
structure
Tree size
Camponotus senex
Pseudomyrmex gracilis
Cephalotes pusillus
26.24* (small)
Camponotus bonariensis
26.27* (small)
Camponotus atriceps
21.86** (large)
Tapinoma sp. 1
Azteca sp. 1
13.14* (SA)
Camponotus melanoticus
Pseudomyrmex curacaensis
12.97* (closed)
Camponotus sericeiventris
Pseudomyrmex elongatus
Solenopsis sp. 1
Crematogasters ampla
11.83*** (small)
Pseudomyrmex urbanus
Between brackets are the habitat characteristics most associated with a given ant species (numbers with
*indicates values of p < .05, with **p < .01 and with ***p < .005). Empty spaces mean that there was not
a significant relationship between a given ant species and a habitat characteristic.
here (Table 3). Similarly, the habitat-­constrained models showed ran-
models became significant with the habitat-­constrained models
dom patterns of co-­occurrence (Table 3). When we considered only
(Table S2).
the 14 most common ant species in the community, the two indexes
showed a significantly segregated pattern of species associations
(Table 3).
3.2.2 | Pairwise-­level co-­occurrence analyses
3.2.3 | Differences between segregated and
aggregated pairs
We found a significant difference in the mean distance in trait dissimilarities between the segregated and aggregated pairs (t = 1.97,
Our pairwise analyses of the 14 most frequent species showed that
p < .05, N = 16). The segregated pairs had a lower mean dissimilarity
21% of the pairs analyzed (19 of 91) have significantly nonrandom
than the segregated ones (Figure 2).
co-­occurrence patterns under either the constrained or uncon-
Of the ten pairs that were segregated under both the uncon-
strained null models (Figure 1a). Of these nonrandom pairwise spe-
strained and constrained models, both species of the pair tended to
cies associations from the unconstrained model analyses, 11 were
have similar nesting/foraging ecology (Table S3). For instance, both
segregated and five were aggregated, while three become nonsig-
Azteca sp. 1 and Crematogaster ampla forage during the day and the
nificant under one or more constrained models, indicative of the as-
night, nest in live and dead branches of any size, recruit massively to
sociations being habitat-­constrained (Figure 1b). Finally, only one of
baits, and have extensive cavity use and large colonies (Tables 1 and
the species pairs that were not significant under the unconstrained
S3). Similarly, Pseudomyrmex gracilis and Pseudomyrmex urbanus, which
T A B L E 3 Co-­occurrence patterns at the
community level of arboreal ant under
unconstrained and habitat-­constrained
analyses
Habitat-­constrained
Species matrix
Unconstrained
Tree species
Tree size
Vegetation
structure
−0.93
−1.11
−1.06
−1.03
0.61
0.32
0.34
0.79
(a) All 75 ant species
SES of St. C-­score
SES of Sorensen
(b) The 14 most frequent ant species
SES of St. C-­score
SES of Sorensen
4.27*
4.21*
4.28*
4.2*
−3.87*
−3.65*
−3.75*
−3.67*
Negative values of standardized effect size (SES) indicate aggregation between species pairs under the
St. C-­score and segregation under the Sorensen index. Positive values of the St. C-­score and negative
values of the Sorensen index indicate segregation between species pairs (SES in bold and with * indicates values of p < .05).
|
CAMAROTA et al.
8 (a)
species were considered, but a significant segregated pattern when
(b)
using only the 14 most common ant species. Moreover, the pairwise
analyses indicated a significant number of segregated and aggregated
11
Segregated
species associations among these most common ant species. Few of
the significant segregated associations became nonsignificant when
Random
72
we took into account the species, size, or the connectivity of the host
tree. Concordantly, few species showed significant associations with a
given tree species, or with trees of a given size or vegetation structure.
5
Nonrandom
Aggregated
Overall, our results suggest that these segregated associations are
more consistent with competitive interactions between species driving species co-­occurrence patterns, with a limited influence of habitat
19
3
Habitat constrained
selection. This result was further supported by our comparisons of the
nesting and foraging ecology of the species that formed segregated
FIGURE 1 Number of ant species associations in pairwise analyses of
the 14 most frequent species in the focal community. (a) Random and
nonrandom associations and (b) within the nonrandom associations
the segregated, aggregated, and habitat-­constrained associations
pairs. Species that formed these pairs almost always shared the same
nesting ecology, foraging ecology, or both, whereas those that formed
aggregated pairs did not.
4.1 | Matrix-­versus pairwise-­level analyses
0.7
The matrix-­level co-­occurrence analyses failed to detect any pattern
0.6
of species association when we accounted for all 75 ant species in the
0.5
ing local ant co-­occurrence patterns at the community level in natural
Trait dissimilarity
community. These results are concordant with other studies assesshabitats, especially recent ones using more robust null models (e.g.,
0.4
Campbell, Fellowes, & Cook, 2015; Gotelli & Ellison, 2002; Sanders,
0.3
ses represent an average of values calculated for all individual pairs of
0.2
the studied community may be obscured in analyses with many spe-
0.1
weaker ones and even cancel each other out if both positive and neg-
Gotelli et al., 2007; Stuble et al., 2013). However, matrix-­level analyspecies (Gotelli, 2000). Therefore, important species interactions in
cies (Gotelli & Ulrich, 2012), as strong interactions can be diluted by
ative interactions exist (Diamond & Gilpin, 1982; Sanderson, 2004).
0.0
This may help to explain why contrasting results were obtained in the
Aggregated
Segregated
Species pairwise association
F I G U R E 2 Mean trait dissimilarity (Sørensen ± SE) between
aggregated and segregated species pairwise associations
analyses involving the whole community and those using only the 14
most common ant species. In the latter case, strong evidence of a segregated co-­occurrence pattern was obtained.
Furthermore, with the pairwise analyses, we found a relatively
large number (~21%) of nonrandom species associations, either segregated or aggregated, thus well above the number expected to occur
also had a significantly segregated co-­occurrence pattern, are strictly
by chance alone (5%). In an analysis of species-­pairs associations from
diurnal species that only nest in small and medium branches. In con-
30 published matrices, Sfenthourakis et al. (2006) showed that only
trast, species that formed aggregated pairs rarely shared the same nest-
eight matrices had a number of deviations (segregation or aggrega-
ing or foraging characteristics (Table S3).
tions) higher than 5%, with just three matrices having more than 10%.
Our results and those of Sfenthourakis et al. (2006) clearly indicate the
need of both matrix-­ and pairwise-­level analyses to the assessment of
4 | DISCUSSION
species associations in a community. Moreover, it also suggests the
We assessed the potential role of interspecific competition and habi-
assessment of already published arboreal ant co-­occurrence matrices.
possibility of novel results with the use of a pairwise approach in a retat selection on the organization of an arboreal ant community in a
Neotropical savanna. We did this by taking advantage of a dataset that
took into account the characteristics of the host trees and abundance
4.2 | Habitat attributes
of ant species in the community. The matrix-­level co-­occurrence
The detection of a significant species association has been fre-
analyses showed random pattern of species associations when all ant
quently used as sufficient evidence to infer a specific ecological
|
9
CAMAROTA et al.
process, and especially competition. Nevertheless, different pro-
In fact, subordinate species have been shown to be better adapted
cesses may produce the same pattern, leaving these inferences
than the dominant ones for foraging at extreme temperatures, avoid-
open to criticism (Gotelli & Graves, 1996; Strong, Simberloff, Abele,
ing direct interactions with the dominant species and the possible
& Thistle, 1984). The structured nature of our dataset allowed us
exclusion from a habitat patch (Bestelmeyer, 2000). This pattern has
to take the next step and evaluate the relative importance of host
been rarely tested in tropical habitats (Wittman et al., 2010), but our
tree selection in explaining the observed patterns. Here, we have
results suggest that it may also be important in tropical arboreal ant
found that when information about the host trees was taken into
communities.
account, the results of our analyses changed very little. In fact, the
Ant recruitment rates also appear to be important in explaining
significance of the observed matrix-­wide analyses did not change
the directions of interspecific associations. More specifically, the re-
with the inclusion of host tree attributes for the 14 most common
cruitment rates were almost always different between coexisting (i.e.,
ant species. Moreover, only a small proportion (15.8%) of the sig-
aggregated) species, which can mean that there is a trade-­off between
nificant pairwise species associations became nonsignificant with
these ants on the ability to find new resource and to dominate or even
the inclusion of host tree characteristics. This indicates that, for ar-
monopolize them. In this trade-­off, also known as “discovery domi-
boreal ants, species aggregations are rarely due to shared host tree
nance,” ants usually differ on their recruitment ability, with superior
preferences. Similarly, in most cases, species segregations could not
competitors having lower recruitment rates and inferior ones investing
be attributed to distinct host tree preferences between ant species.
all the energy of the colony in finding resources (Fellers, 1987; Parr &
These results probably reflect the fact that none of the most com-
Gibb, 2012). The extent to which these kinds of trade-­offs are import-
mon species in our community showed strong affinities for trees
ant in tropical arboreal ant communities can therefore be seen as a
from a given vegetation structure, size or species (Table 2). Our re-
valuable focus for future work.
sults contrast with those of Azeria et al. (2012) who showed that
Despite focusing on different aspects of the foraging habits of dif-
co-­occurrence patterns of saproxylic beetles are largely modulated
ferent ant species, it is important to note that we did not take into ac-
by the species and size of the host tree, rather than by interspecific
count dietary preferences of the different ant species as an important
competition.
trait to explain ant species coexistence. However, most arboreal ants
have diets based on sugar-­rich exudates (i.e., honeydew and extra-
4.3 | Species traits
floral nectar), acting as functional herbivores, with very few cases of
true predators (Blüthgen, Gebauer, & Fiedler, 2003; Davidson, Cook,
Broadly, our results indicated a number of significant positive and
Snelling, & Chua, 2003; Russell et al., 2009). Moreover, the presence
negative pairwise species associations, but with no community-­wide
of food seems to not be a major axis of niche differentiation, with only
signature and no real influence of host tree attributes. The significant
limited effects on the structure of arboreal ant communities in the
associations are therefore better explained by competition among
focal habitat of this study (Camarota, Powell, Vasconcelos, Priest, &
specific pairs of species, and we may need to turn to the traits of the
Marquis, 2015).
ants as an explanation. Indeed, our data give support to the prediction
that trait-­mediated competitive processes are important in explaining
interspecific associations (Adler et al., 2014; Chesson, 2000), as the
co-­occurring species (aggregated) had fewer similarities in common
than the segregated species.
4.4 | Competition in arboreal ant communities
versus the ant mosaic hypothesis
Ant communities are often thought to be structured by interspecific
Nesting sites can be a limited resource for cavity-­nesting arboreal
competition (Hölldobler & Wilson, 1990; Parr & Gibb, 2010), but many
ants (Blüthgen & Feldhaar, 2010; Philpott & Foster, 2005; Powell et al.,
assumptions allowing for such interpretation still need to be properly
2011), with many ant species showing some level of specialization
tested (Cerdá, Arnan, & Retana, 2013). One of these assumptions
over the use of these resources (Powell et al., 2011). Our data showed
states that arboreal ant communities are spatially distributed in a mo-
that more than half of the segregated pairs of species have marked
saic formed by the mutually exclusive foraging territories of dominant
similarities in their nesting habits, and this similarity can be related
ant species (Leston, 1970; Majer, 1972; Room, 1971). In the classi-
either to the structure and/or the extensive use of their nests. In con-
cal examples of ant mosaic, the competition over territory is limited
trast, all coexisting species pairs had markedly different nesting habits.
to behaviorally dominant species with large colonies (Bluthgen &
Campbell, Fellowes, and Cook (2013) also showed the importance of
Stork, 2007). In our study system, only two species (Azteca sp. 1 and
nest site selection on domatia-­dwelling ant species coexistence.
Crematogaster ampla) fulfilled this criterion of dominance and never
Another important aspect in defining the directions of interspe-
co-­occurred on the same individual tree. It is interesting to note that
cific associations was the activity period of the ant species. Almost
these two species are ecologically very similar, using similar nesting
half of the negatively associated pairs were formed by species forag-
sites, occupying many nests per colony, foraging during both day and
ing on the same time schedule. Indeed, there are a number of studies
night, and have massive recruitment capacity and small individual
showing the importance of activity time for ant species coexistence
body size. Moreover, these ant genera (Azteca and Crematogaster)
(Cerdá, Retana, & Cros, 1997; Cerdá, Retana, & Manzaneda, 1998;
are characterized by the presence of a modified proventriculus that
Lessard, Dunn, & Sanders, 2009; Stringer, Haywood, & Lester, 2007).
enables them to harvest liquid food resources in a very effective way,
|
CAMAROTA et al.
10 reinforcing their dominance status (Davidson, 1997, 2005; Davidson,
Cook, & Snelling, 2004).
The ant mosaic hypothesis is a specific predicated pattern of spatial structuring that emerges from competition over food in arboreal
ant communities. Direct tests of this predicted pattern of competition-­
driven spatial structuring have dominated the literature on community assembly in arboreal ant communities for many years (Blüthgen
& Stork 2007; Jackson, 1984; Room, 1971). The central part of this
hypothesis is that the whole habitat would be divided up into large
mutually exclusive territories (Leston, 1973). We did not observe this
pattern in our study. Moreover, one of the other expected patterns of
the ant mosaic hypothesis is a well-­defined set of subdominant and
subordinate ant species associated with each particular dominant ant
species (Blüthgen et al. 2007; Majer, Delabie, & Smith, 1994; Room,
1971). This was also not seen in our analyses, with no detection of
a particular set of species related to a given dominant ant species.
Overall, our data showed support for a pattern of nonrandom spatial
structuring, consistent with an important role of competition, but no
evidence supporting the specific spatial structuring predicted by the
ant mosaic hypotheses. These findings therefore advocate for future
work to address the broader and more nuanced influences of competition on the spatial structuring of arboreal ant communities.
5 | CONCLUSION
While our matrix-­level co-­occurrence analyses showed a random pattern of species associations with all ant species considered, we found
a significant segregated pattern when using only the most common
species. In addition, the pairwise analyses identified a significant
number of nonrandom species associations among these most common ant species. Overall, our results suggest that these segregated
associations are more consistent with competitive interactions between species driving species co-­occurrence patterns, with a limited
influence of habitat selection. This result was further supported by
our comparisons of important traits of the ant species that formed
segregated pairs. Therefore, these findings provide support for an
important role for trait-­mediated competitive interactions in determining coexistence in the focal arboreal ant community. Our analyses
were unusual in the extent to which they were able to integrate a
suite of potentially important environmental influences on assembly,
providing greater certainty in interpreting the processes underlying
the patterns we identified. We detected frequent segregation of the
most common ant species. Moreover, despite the majority of the species pairs not forming significant associations and the lack of strong
signs of habitat choices, we could find strong associations between
biological characteristics and the significantly segregated or aggregated species pairs. Broadly, our findings suggest that trait-­mediated
competitive interactions have more nuanced outcomes and spatial
structuring in arboreal ant communities than has been previously considered. They also highlight the value of biodiversity inventory studies
that explicitly incorporate potentially important environmental influences on assembly, and analyses that integrate such data.
AC KNOW L ED G M ENTS
The authors would like to thank M. Gonzaga, R. Feitosa, R. Pacheco
and S. Sendoya for comments that improved this manuscript.
This work was funded by the National Science Foundation (DEB
0842144) and the Brazilian Council for Research and Scientific
Development (CNPq grants 478107/2012-­9 and 457407/2012-­3).
S. Powell received additional support from National Science
Foundation grant IOS 0841756 and research funds from the George
Washington University.
CO NFL I C T O F I NT ER ES T
None declared.
REFERENCES
Adler, P. B., Salguero-Gómez, R., Compagnoni, A., Hsu, J. S., Ray-Mukherjee,
J., Mbeau-Ache, C., & Franco, M. (2014). Functional traits explain variation in plant life history strategies. Proceedings of the National Academy
of Sciences, 111, 740–745.
Agrawal, A. A., Ackerly, D. D., Adler, F., Arnold, A. E., Cáceres, C., Doak, D.
F., ··· Werner, E. (2007). Filling key gaps in population and community
ecology. Frontiers in Ecology and the Environment, 5, 145–152.
Azeria, E. T., Fortin, D., Hébert, C., Peres-Neto, P., Pothier, D., & Ruel, J.-C.
(2009). Using null model analysis of species co-­occurrences to deconstruct biodiversity patterns and select indicator species. Diversity and
Distributions, 15, 958–971.
Azeria, E., Ibarzabal, J., & Hébert, C. (2012). Effects of habitat characteristics and interspecific interactions on co-­occurrence patterns of saproxylic beetles breeding in tree boles after forest fire: Null model analyses.
Oecologia, 168, 1123–1135.
Bestelmeyer, B. T. (2000). The trade-­off between thermal tolerance and behavioural dominance in a subtropical South American ant community.
Journal of Animal Ecology, 69, 998–1009.
Blüthgen, N., & Feldhaar, H. (2010). Food and shelter: How resources influence ant ecology. In L. Lach, C. L. Parr & K. L. Abbott (Eds.), Ant ecology,
1st ed. (pp. 115–136). Oxford: Oxford University Press.
Blüthgen, N., Gebauer, G., & Fiedler, K. (2003). Disentangling a rainforest
food web using stable isotopes: Dietary diversity in a species-­rich ant
community. Oecologia, 137, 426–435.
Blüthgen, N., & Stork, N. E. (2007). Ant mosaics in a tropical rainforest in
Australia and elsewhere: A critical review. Austral Ecology, 32, 93–104.
Cáceres, M. D., & Legendre, P. (2009). Associations between species
and groups of sites: Indices and statistical inference. Ecology, 90,
3566–3574.
Camarota, F., Powell, S., Vasconcelos, H. L., Priest, G., & Marquis, R. J. (2015).
Extrafloral nectaries have a limited effect on the structure of arboreal
ant communities in a Neotropical savanna. Ecology, 96, 231–240.
Campbell, H., Fellowes, M. D. E., & Cook, J. M. (2013). Arboreal thorn-­
dwelling ants coexisting on the savannah ant-­plant, Vachellia erioloba,
use domatia morphology to select nest sites. Insectes sociaux, 60,
373–382.
Campbell, H., Fellowes, M. D., & Cook, J. M. (2015). The curious case of
the camelthorn: Competition, coexistence, and nest-­site limitation in
a multispecies mutualism. The American Naturalist, 186, E172–E181.
Campos, R. I., Vasconcelos, H. L., Ribeiro, S. P., Neves, F. S., & Soares, J. P.
(2006). Relationship between tree size and insect assemblages associated with Anadenanthera macrocarpa. Ecography, 29, 442–450.
Cerdá, X., Arnan, X., & Retana, J. (2013). Is competition a significant hallmark of ant (Hymenoptera: Formicidae) ecology? Myrmecological News,
18, 131–147.
CAMAROTA et al.
Cerdá, X., Retana, J., & Cros, S. (1997). Thermal disruption of transitive hierarchies in mediterranean ant communities. Journal of Animal Ecology,
66, 363–374.
Cerdá, X., Retana, J., & Manzaneda, A. (1998). The role of competition by
dominants and temperature in the foraging of subordinate species in
mediterranean ant communities. Oecologia, 117, 404–412.
Chase, J. M., & Leibold, M. A. (2003). Ecological niches. Chicago, IL:
University of Chicago Press.
Chesson, P. (2000). Mechanisms of maintenance of species diversity.
Annual Review of Ecology and Systematics, 31, 343–358.
Connor, E. F., & Simberloff, D. (1979). The assembly of species communities: Chance or competition? Ecology, 60, 1132–1140.
Connor, E. F., & Simberloff, D. (1984). Neutral models of species’ co-occurrence patterns. In D. R. Strong, D. Simberloff, L. G. Abele & A. B. Thistle
(Eds.), Ecological communities: Conceptual issues and the evidence (pp.
316–331). Princeton, NJ: Princeton University Press.
Davidson, D. W. (1997). The role of resource imbalances in the evolutionary ecology of tropical arboreal ants. Biological Journal of the Linnean
Society, 61, 153–181.
Davidson, D. (2005). Ecological stoichiometry of ants in a New World rain
forest. Oecologia, 142, 221–231.
Davidson, D., Cook, S., & Snelling, R. (2004). Liquid-­feeding performances of ants (Formicidae): Ecological and evolutionary implications.
Oecologia, 139, 255–266.
Davidson, D. W., Cook, S. C., Snelling, R. R., & Chua, T. H. (2003). Explaining
the abundance of ants in lowland tropical rainforest canopies. Science,
300, 969–972.
Dejean, A., Fisher, B. L., Corbara, B., Rarevohitra, R., Randrianaivo, R.,
Rajemison, B., & Leponce, M. (2010). Spatial distribution of dominant
arboreal ants in a malagasy coastal rainforest: Gaps and presence of an
invasive species. PLoS One, 5, e9319.
Diamond, J. (1975). Assembly of species communities. In M. L. Cody & J.
M. Diamond (Eds.), Ecology and evolution of communities (pp. 342–344).
Cambridge, MA: Harvard University Press.
Diamond, J., & Gilpin, M. (1982). Examination of the “null” model of Connor
and Simberloff for species co-­occurrences on Islands. Oecologia, 52,
64–74.
Dice, L. R. (1945). Measures of the amount of ecologic association between
species. Ecology, 26, 297–302.
Djiéto-Lordon, C., Dejean, A., Gibernau, M., Hossaert-McKey, M., & McKey,
D. (2004). Symbiotic mutualism with a community of opportunistic
ants: Protection, competition, and ant occupancy of the myrmecophyte
Barteria nigritana (Passifloraceae). Acta Oecologica, 26, 109–116.
Dufrene, M., & Legendre, P. (1997). Species assemblages and indicator
species: The need for a flexible asymmetrical approach. Ecological
Monographs, 67, 345–366.
Fellers, J. H. (1987). Interference and exploitation in a guild of woodland
ants. Ecology, 68, 1466–1478.
Fowler, D., Lessard, J.-P., & Sanders, N. J. (2014). Niche filtering rather than
partitioning shapes the structure of temperate forest ant communities.
Journal of Animal Ecology, 83, 943–952.
Gotelli, N. J. (2000). Null model analysis of species co-­occurrence patterns.
Ecology, 81, 2606–2621.
Gotelli, N. J. (2001). Research frontiers in null model analysis. Global Ecology
& Biogeography, 10, 337–343.
Gotelli, N. J., & Ellison, A. M. (2002). Assembly rules for New England ant
assemblages. Oikos, 99, 591–599.
Gotelli, N. J., & Entsminger, G. (2001). Swap and fill algorithms in null model
analysis: Rethinking the knight’s tour. Oecologia, 129, 281–291.
Gotelli, N. J., & Entsminger, G. L. (2003). Swap algorithms in null model
analysis. Ecology, 84, 532–535.
Gotelli, N. J., & Graves, G. R. (1996). Null models in ecology: Linking classic
and contemporary approaches. Washington, DC: Smithsonian Institution
Press.
|
11
Gotelli, N. J., & McCabe, D. J. (2002). Species co-­occurrence: A meta-­analysis
of j. m. diamond’s assembly rules model. Ecology, 83, 2091–2096.
Gotelli, N. J., & Ulrich, W. (2012). Statistical challenges in null model analysis. Oikos, 121, 171–180.
Hölldobler, B., & Wilson, E. (1990). The ants. Cambridge, MA: Harvard
University Press.
Hutchinson, G. E. (1959). Homage to Santa-­Rosalia or why are there so
many kinds of animals. American Naturalist, 93, 145–159.
Jackson, D. A. (1984). Ant distribution patterns in a cameroonian cocoa
plantation: Investigation of the ant mosaic hypothesis. Oecologia, 62,
318–324.
Klimes, P., Idigel, C., Rimandai, M., Fayle, T. M., Janda, M., Weiblen, G. D.,
& Novotny, V. (2012). Why are there more arboreal ant species in primary than in secondary tropical forests? Journal of Animal Ecology, 81,
1103–1112.
Koch, E. B. A., Camarota, F., & Vasconcelos, H. L. (2016). Plant ontogeny as
a conditionality factor in the protective effect of ants on a neotropical
tree. Biotropica, 48, 198–205.
Lach, L., Parr, C. L., & Abbot, K. L. (2010). Ant ecology. Oxford: Oxford
University Press.
Lessard, J. P., Dunn, R., & Sanders, N. (2009). Temperature-­mediated coexistence in temperate forest ant communities. Insectes Sociaux, 56,
149–156.
Leston, D. (1970). Entomology of cocoa farm. Annual Review of Entomology,
15, 273–294.
Leston, D. (1973). The ant mosaic-­tropical tree crops and the limiting of
pests and diseases. PANS Pest Articles & News Summaries, 19, 311–341.
Levine, J. M., & HilleRisLambers, J. (2009). The importance of niches for the
maintenance of species diversity. Nature, 461, 254–257.
MacArthur, R., & Levins, R. (1967). The limiting similarity, convergence,
and divergence of coexisting species. The American Naturalist, 101,
377–385.
Majer, J. D. (1972). The ant mosaic in Ghana cocoa farms. Bulletin of
Entomological Research, 62, 151–160.
Majer, J. D., Delabie, J. H. C., & Smith, M. R. B. (1994). Arboreal ant community patterns in Brazilian cocoa farms. Biotropica, 26, 73–83.
Moreno, M. I. C. (2005). Estado nutricional de espécies lenhosas e disponibilidade de nutrientes no solo e na serapilheira em diferentes fitofisionomias do cerrado na região do triângulo mineiro, Master Thesis.
Universidade de Brasília, Brasília.
Myers, N., Mittermeier, R. A., Mittermeier, C. G., Da Fonseca, G. A., & Kent,
J. (2000). Biodiversity hotspots for conservation priorities. Nature, 403,
853–858.
Oksanen, J., Kindt, R., Legendre, P., O’Hara, B., & Stevens, M. H. H. (2013).
VEGAN: Community ecology package (R package version 3.3-3). Vienna,
Austria: R Foundation for Statistical Computing.
Oliveira-Filho, A. T., & Ratter, J. A. (2002). Vegetation physiognomies and
woody flora of the cerrado biome. In P. S. Oliveira & R. J. Marquis
(Eds.), The Cerrados of Brazil: Ecology and Natural History of a Neotropical
Savanna (pp.91–120). New York: Columbia University Press.
Pacheco, R., & Vasconcelos, H. L. (2012). Habitat diversity enhances ant
diversity in a naturally heterogeneous Brazilian landscape. Biodiversity
and Conservation, 21, 797–809.
Parr, C. L., & Gibb, H. (2010). Competition and the role of dominant ants. In
L. Lach, C. L. Parr & K. L. Abbot (Eds.), Ant ecology (pp. 77–96). Oxford:
Oxford University Press.
Parr, C. L., & Gibb, H. (2012). The discovery–dominance trade-­off is the
exception, rather than the rule. Journal of Animal Ecology, 81, 233–241.
Peres-Neto, P. R., Olden, J. D., & Jackson, D. A. (2001). Environmentally
constrained null models: Site suitability as occupancy criterion. Oikos,
93, 110–120.
Pfeiffer, M., Cheng Tuck, H., & Chong Lay, T. (2008). Exploring arboreal ant community composition and co-­occurrence patterns in plantations of oil palm
Elaeis guineensis in Borneo and Peninsular Malaysia. Ecography, 31, 21–32.
|
CAMAROTA et al.
12 Philpott, S. M., & Foster, P. F. (2005). Nest-­site limitation in coffee agroecosystems: Artificial nests maintain diversity of arboreal ants. Ecological
Applications, 15, 1478–1485.
Pitta, E., Giokas, S., & Sfenthourakis, S. (2012). Significant pairwise co-­
occurrence patterns are not the rule in the majority of biotic communities. Diversity, 4, 179–193.
Powell, S., Costa, A. N., Lopes, C. T., & Vasconcelos, H. L. (2011). Canopy
connectivity and the availability of diverse nesting resources affect species coexistence in arboreal ants. Journal of Animal Ecology, 80, 352–360.
R Development Core Team (2014). R: A language and environment for statistical computing. R foundation for statistical computing. Retrieved
from http://www.r-project.org
Ribas, C. R., & Schoereder, J. H. (2002). Are all ant mosaics caused by competition?. Oecologia, 131, 606–611.
Ribas, C. R., Schoereder, J. H., Pic, M., & Soares, S. M. (2003). Tree heterogeneity, resource availability, and larger scale processes regulating
arboreal ant species richness. Austral Ecology, 28, 305–314.
Room, P. M. (1971). The relative distributions of ant species in ghana’s
cocoa farms. Journal of Animal Ecology, 40, 735–751.
Russell, J. A., Moreau, C. S., Goldman-Huertas, B., Fujiwara, M., Lohman, D.
J., & Pierce, N. E. (2009). Bacterial gut symbionts are tightly linked with
the evolution of herbivory in ants. Proceedings of the National Academy
of Sciences, 106, 21236–21241.
Sanders, N. J., Crutsinger, G. M., Dunn, R. R., Majer, J. D., & Delabie, J. H. C.
(2007). An ant mosaic revisited: Dominant ant species disassemble arboreal ant communities but co-­occur randomly. Biotropica, 39, 422–427.
Sanders, N. J., Gotelli, N. J., Wittman, S. E., Ratchford, J. S., Ellison, A. M., &
Jules, E. S. (2007). Assembly rules of ground-­foraging ant assemblages
are contingent on disturbance, habitat and spatial scale. Journal of
Biogeography, 34, 1632–1641.
Sanderson, J. G. (2004). Null model analysis of communities on gradients.
Journal of Biogeography, 31, 879–883.
Schoener, T. W. (1974). Resource partitioning in ecological communities.
Science, 185, 27–39.
Sfenthourakis, S., Tzanatos, E., & Giokas, S. (2006). Species co-­occurrence:
The case of congeneric species and a causal approach to patterns of
species association. Global Ecology and Biogeography, 15, 39–49.
Stone, L., & Robert, A. (1990). The checkerboard score and species distributions. Oecologia, 85, 74–79.
Stringer, L. D., Haywood, J., & Lester, P. J. (2007). The influence of temperature and fine-­scale resource distribution on resource sharing
and domination in an ant community. Ecological Entomology, 32,
732–740.
Strong, D. R., Simberloff, D., Abele, L. G., & Thistle, A. B. (1984). Ecological
communities: Conceptual issues and the evidence. Princeton, NJ:
Princeton University Press.
Stuble, K. L., Rodriguez-Cabal, M. A., McCormick, G. L., Jurić, I., Dunn,
R. R., & Sanders, N. J. (2013). Tradeoffs, competition, and coexistence in eastern deciduous forest ant communities. Oecologia, 171,
981–992.
Sutherland, W. J., Freckleton, R. P., Godfray, H. C. J., Beissinger, S. R.,
Benton, T., Cameron, D. D., ··· Wiegand, T. (2013). Identification of 100
fundamental ecological questions. Journal of Ecology, 101, 58–67.
Ulrich, W., & Gotelli, N. J. (2013). Pattern detection in null model analysis.
Oikos, 122, 2–18.
Veech, J. A. (2014). The pairwise approach to analysing species co-­
occurrence. Journal of Biogeography, 41, 1029–1035.
Wittman, S. E., Sanders, N. J., Ellison, A. M., Jules, E. S., Ratchford, J. S., &
Gotelli, N. J. (2010). Species interactions and thermal constraints on
ant community structure. Oikos, 119, 551–559.
S U P P O RT I NG I NFO R M AT I O N
Additional Supporting Information may be found online in the supporting information tab for this article.
How to cite this article: Camarota, F., Powell, S., S. Melo, A., Priest,
G., J. Marquis, R. and L. Vasconcelos, H. (2016), Co-­occurrence
patterns in a diverse arboreal ant community are explained more by
competition than habitat requirements. Ecology and Evolution,
00:1–12. doi: 10.1002/ece3.2606